| *** HarryHaaren has quit IRC | 00:48 | |
| drobilla | Blergh, zero motivation today. | 01:14 |
|---|---|---|
| drobilla | Maybe I'll just commit the show()/hide() interface part and leave it at that | 01:14 |
| drobilla | 10 Internet Points to someone who actually wants to make a UI that uses it :) | 01:15 |
| drobilla | Super annoying that run() was the first method in lv2_external_ui, or it could (perhaps) conveniently be castable | 01:15 |
| drobilla | Well, it uses a different 'this' pattern than LV2 stuff anyway, I guess, so whatever | 01:16 |
| * drobilla is having one of those days where you should have just accepted from the outset that you weren't going to get anything done and just gone with it | 01:18 | |
| rgareus | drobilla: nice job on the 5th plugin | 02:46 |
| rgareus | drobilla: did you check what happens if you send a note with pitch > 120 ? does it wrap? | 02:46 |
| drobilla | rgareus: It just doesn't bother sending a fifth | 02:46 |
| rgareus | drobilla: fifth.msg[1] = msg[1] + 7; // Pitch up 7 semitones | 02:47 |
| drobilla | rgareus: Did I actually test this? no. :) | 02:47 |
| drobilla | rgareus: if (note < 127 - 7) { | 02:47 |
| rgareus | msg is a uint8_t so it will produce invalid midi-data, value > 127 | 02:47 |
| rgareus | (or a &0x7f | 02:47 |
| rgareus | = (msg[1] + 7) & 0x7f; | 02:47 |
| rgareus | drobilla: small detail really. but producing illegal midi-data shoud be discouraged. | 02:48 |
| drobilla | ... | 02:49 |
| drobilla | See above | 02:49 |
| * rgareus head -> keyboad | 02:50 | |
| drobilla | :) | 02:51 |
| rgareus | mmh. that should be <= 127 - 7 or < 128 - 7 | 02:51 |
| drobilla | true | 02:51 |
| rgareus | and it does not explain how to come up with an event from scratch. here: fifth.event = *ev; | 02:53 |
| rgareus | take care of this | 02:53 |
| drobilla | meh. | 02:53 |
| rgareus | drobilla: well I'm just reviewing code. nothing personal. | 02:53 |
| drobilla | I guess since it's an example that rather clunky might be appropriate | 02:54 |
| drobilla | rgareus: Appreciated. meh isn't personal. Go fuck yourself is personal :) | 02:54 |
| drobilla | Apathy: Where everyone gets along because the alternative is just too much bother™ | 02:54 |
| drobilla | This 64 bit time stamp thing was pretty silly. Oh well. | 02:55 |
| rgareus | right. good enough for an example. and events are well documented at http://lv2plug.in/doc/html/structLV2__Atom__Event.html | 02:56 |
| rgareus | 2^32 is not enough if I want to feed 1 minute of intersetall communication sampled at 72MSPS into my marscomunicationdecoder.lv2 | 03:00 |
| drobilla | rgareus: Heh. Well, there was 64-bits regardless for alignment | 03:00 |
| rgareus | s/intersetall/interstellar/ | 03:00 |
| drobilla | frames and subframes ala even may have been more appropriate | 03:00 |
| drobilla | ala event* | 03:00 |
| drobilla | Get signed to unsigned conversion errors with hyper warnings on | 03:01 |
| drobilla | The fifth thing was rev 909, heh | 03:17 |
| drobilla | Should have made it a beat repeat or something :) | 03:17 |
| drobilla | rgareus: thanks http://lv2plug.in/changeset/910 | 03:17 |
| rgareus | drobilla: nice. | 03:18 |
| drobilla | I'm too lazy to write a UI | 03:21 |
| drobilla | Velocity scale of the fifth is the obvious control, I suppose | 03:22 |
| rgareus | it does not need one, does it? | 03:22 |
| rgareus | even velocity scale can get away with a simple float control port | 03:22 |
| drobilla | Need one what? | 03:24 |
| rgareus | UI | 03:24 |
| drobilla | Well, no. Few things do ;) | 03:24 |
| rgareus | drobilla: what was your comment "I'm too lazy to write a UI" about, then? | 03:24 |
| drobilla | I need a simple preferably pugl example to do the external degrade thing | 03:24 |
| drobilla | Hopefully in the process winning my freedom from that shit for as long as possible. | 03:24 |
| rgareus | drobilla: an on screen keyboard | 03:25 |
| drobilla | I'm /way/ too lazy for that | 03:25 |
| rgareus | drobilla: and degration would be hard | 03:25 |
| drobilla | Why? | 03:25 |
| rgareus | drobilla: what would it degrate to? | 03:25 |
| drobilla | rgareus: external UI | 03:26 |
| drobilla | via an additional interface for that | 03:27 |
| rgareus | in libsuil? | 03:27 |
| drobilla | as discussed in my ML post nobody contributes to and yammered on about here in the past several days | 03:27 |
| drobilla | Since there seems to be no reasonable way to not give a host the widget it's expecting, probably not anything in suil at all, no. | 03:29 |
| rgareus | I don't yet have anything constructive to contribute to that thread (much like Harry). | 03:30 |
| drobilla | Then don't make me repeat myself on IRC :P | 03:31 |
| rgareus | drobilla: it wasn't clear to me that you planned to fall back to externalUI. | 03:31 |
| drobilla | In operation, not literally. | 03:31 |
| drobilla | The whole reason external UI is a huge problem is it is a different UI type, to the exclusion of others. | 03:32 |
| drobilla | The very opposite of graceful, and a big clusterfuck of fragmentation. | 03:32 |
| drobilla | I haven't yet thought of any reason why putting the same stuff in an extension_data interface isn't just completely better in every way | 03:33 |
| drobilla | Host doesn't know how to deal with a WtfTkUI? Okay, no embedding for you, call show() on it. | 03:34 |
| rgareus | the UI type of external UI is toolkit independent. What is the problem with that? | 03:36 |
| drobilla | The problem is that it's a separate UI type | 03:36 |
| drobilla | Which just causes pointless fragmentation, wars about embedding, hosts and UIs clashing with each other over what's supported | 03:37 |
| drobilla | You yourself were complaining about having to provide a bunch of different UI types | 03:37 |
| drobilla | The separate UI mechanism was intended for actually different UIs | 03:37 |
| drobilla | Not 40 versions of the same damned thing in a slightly different form for compatibility, which is clearly shit | 03:37 |
| rgareus | Ok I now understand what you meant | 03:37 |
| drobilla | The other part, allowing UIs to provide multiple types (maybe Gtk or X11, or even X11 or Wayland) is more questionble | 03:39 |
| rgareus | speaking of different UI types; could there be a UI:PuGL type (that is identical on all platforms, libpugl chooses the right backend) | 03:40 |
| drobilla | Could there really? Ultimately Pugl just uses magic GL context | 03:40 |
| drobilla | There's not really a meaningful handle of any use to a host. | 03:40 |
| drobilla | and anyway, as discussed yesterday, I think that's just more of the problem. | 03:41 |
| rgareus | currently I use pugl+gtk and in the Makefile use sed to set CocoaUI vs X11UI | 03:41 |
| drobilla | 'system' window handles are the sensible thing | 03:41 |
| drobilla | they don't add library compatibility problems to the host:UI interface | 03:41 |
| rgareus | uhm scratch that +gtk | 03:41 |
| drobilla | yes, the multi thing is an idea to make having to list cocoa, X11, blah blah separately | 03:42 |
| drobilla | All of which is why fifths "needs" a UI | 03:42 |
| rgareus | pugl can work on the three major platforms with different backend, so the host need not really worry | 03:42 |
| drobilla | A *single* pugl UI that works in every single one of the above mentioned scenarios | 03:42 |
| rgareus | drobilla: do you know what harrison does? | 03:43 |
| drobilla | rgareus: Not really. IIRC they've waffled on embedding vs external | 03:43 |
| rgareus | drobilla: they have GL LV2 plugins that work on all 3 platforms | 03:43 |
| rgareus | and there's pugl in there | 03:44 |
| drobilla | rgareus: Not really. As far as LV2 is concerned they have 3 UIs | 03:44 |
| drobilla | I meah, yeah, as far as actually implementing it, pugl works on all 3. That's the whole point of it. | 03:45 |
| drobilla | The mess is at the host:ui interface | 03:45 |
| rgareus | drobilla: and they use externalUI. | 03:46 |
| drobilla | Hooray, no embedding even if the host totally can and wants to | 03:47 |
| drobilla | ^ the problem. | 03:47 |
| rgareus | yeah | 03:47 |
| rgareus | ui:Pugl could facilitat that | 03:48 |
| rgareus | facilitate, even | 03:48 |
| drobilla | I see literally no advantage to doing that and plenty of disadvantages. | 03:48 |
| drobilla | (The main one being a Gtk style library version compatibility clusterfuck) | 03:48 |
| drobilla | What is the hadle for a PuglUI anyway? PuglView? Of no use to the host. | 03:49 |
| drobilla | If the UI is running off system windows events anyway (which is what Pugl does), just... let it. | 03:50 |
| drobilla | GL is weird, for embedding of GL in GL literally we'd need to invent an interface to set up the bounds and projection and whatnot | 03:51 |
| rgareus | indeed. I have not thought about that. | 03:52 |
| rgareus | high time for bed, I suppose. my brain is already sleeping | 03:52 |
| drobilla | Which would be cool, but pragmatic solutions to real problems first :) | 03:52 |
| rgareus | drobilla: speaking of gtk-style. I've fixed the button+dial gradient issues in robtk. (already in meters.lv2, other projects will follow soon) | 03:54 |
| drobilla | rgareus: What meter has widgets? | 03:56 |
| rgareus | drobilla: ebur128, goniometer, phase-wheel, 1/3octave spectrum analyser. | 03:57 |
| rgareus | drobilla: https://github.com/x42/meters.lv2#screenshots | 03:57 |
| drobilla | yeah, buttons in goniometer look ok | 03:58 |
| rgareus | drobilla: strictly speaking the calibration screw is also a widget (at least it's auto-aligned when the UI is resized) | 03:58 |
| * rgareus likes scalable UIs | 04:01 | |
| rgareus | http://robin.linuxaudio.org/tmp/bbcmeter.jpg | 04:01 |
| drobilla | Mm. Audio plugins in general have a severe fixed size problem. | 04:02 |
| * drobilla wonders if easily stealable piano keyboard drawing+event mapping code exists somewhere | 04:03 | |
| rgareus | drobilla: for just black/white keys, it's easier to whip up something from scratch than steal. | 04:04 |
| drobilla | Meh, 11pm, whatever. I need to do some real work anyway | 04:05 |
| * drobilla is trying to retain a vaguely normal human being sleep schedule | 04:05 | |
| rgareus | drobilla: ./gtk2_ardour/gtk_pianokeyboard.c has a gtk+cairo variant. | 04:07 |
| rgareus | the expose_event() and draw_note() is < 100 LOC | 04:07 |
| rgareus | < 50, even | 04:07 |
| drobilla | Cairo pugl is another thing I'd like to do | 04:08 |
| drobilla | Shit, I forgot to try to install OSX today | 04:08 |
| drobilla | Clearly infeasible to maintain Pugl without being able to test myself | 04:08 |
| rgareus | speaking of pugl-merge-mess: one thing that I've changed in pugl is to make puglDisplay() not call glClear() nor glLoadIdentity(); | 04:14 |
| rgareus | that's a show stopper for partial exposure | 04:14 |
| drobilla | It probably just needs to not mess with GL context at all. | 04:14 |
| drobilla | Any GL program is responsbile for that stuff anyway. | 04:15 |
| rgareus | drobilla: +1; but then there needs to be an abstraction for glXMakeCurrent() and similar on other platforms | 04:15 |
| rgareus | anyway. tomorrow. or later | 04:16 |
| drobilla | or pervasive EGL support, which would be nice | 04:17 |
| drobilla | rgareus: Well, not make current, but clear and identity and setting projection and whatnot | 04:17 |
| drobilla | rgareus: 'night | 04:17 |
| *** awilliams has joined #lv2 | 04:22 | |
| *** timbyr has quit IRC | 04:29 | |
| *** timbyr has joined #lv2 | 04:30 | |
| *** Gethiox has quit IRC | 07:10 | |
| *** falktx has joined #lv2 | 07:24 | |
| *** mlpug has joined #lv2 | 08:11 | |
| *** falktx has quit IRC | 09:47 | |
| *** mlpug has quit IRC | 10:08 | |
| *** falktx has joined #lv2 | 11:04 | |
| *** rncbc has joined #lv2 | 12:52 | |
| *** Gethiox has joined #lv2 | 13:00 | |
| *** rncbc has quit IRC | 13:44 | |
| *** HarryHaaren has joined #lv2 | 13:54 | |
| HarryHaaren | +1 for the fifths plugin: its really clear for understanding MIDI through | 14:03 |
| *** falktx has quit IRC | 14:37 | |
| *** mlpug has joined #lv2 | 14:48 | |
| *** falktx has joined #lv2 | 15:04 | |
| *** NickSB2 has quit IRC | 15:10 | |
| rgareus | ls /tmp/jalv-* | wc -l # returns 1285 | 15:29 |
| rgareus | what are all those tmpdirs doing there? | 15:29 |
| rgareus | (they're all empty) | 15:29 |
| *** awilliams has quit IRC | 15:33 | |
| rgareus | does that mean I had 1285 jalv crashes? or more likley gdb jalv -> just quit; sessions since last reboot!? | 15:36 |
| rgareus | does anyone else have a large amount of tmpfiles there? | 15:36 |
| HarryHaaren | rgareus, yep. I guess one tmp dir per instance | 15:37 |
| falktx | ls /tmp/jalv* | wc -l | 15:37 |
| falktx | 3 | 15:37 |
| rgareus | falktx: 3 only? | 15:38 |
| falktx | I only ran jalv a couple of times | 15:38 |
| falktx | running it again increases to 5 | 15:38 |
| falktx | now 7... | 15:38 |
| rgareus | well my uptime is several month. but /tmp is cleaded every reboot or files oder than 1 month. | 15:39 |
| rgareus | falktx: do they go away when you quit jalv? | 15:39 |
| rgareus | they should -- remove(jalv.temp_dir); at the end. | 15:39 |
| falktx | rgareus: they don't if I use ctrl+c | 15:40 |
| falktx | exiting cleanly is ok (count doesn't increase) | 15:40 |
| rgareus | ok. so it was /me using ctrl+c or gdb quit > 1K times. wow | 15:41 |
| falktx | BUT using ctrl+c with an external UI is fine | 15:41 |
| falktx | rgareus: > 500 times. it increases the value by 2 each time | 15:41 |
| *** falktx has quit IRC | 15:44 | |
| rgareus | not here. only one. | 15:44 |
| HarryHaaren | uptime of months? I have a kernel panic every few weeks :D | 15:45 |
| rgareus | HarryHaaren: that's whay I'm still on 3.2 :) | 15:46 |
| rgareus | HarryHaaren: works reliably | 15:47 |
| HarryHaaren | rgareus, I'm planning on buying a new SSD, and doing a performance-only install on it: 3.2 is Debian + custom config, or AvLinux? | 15:48 |
| rgareus | HarryHaaren: it's a custom kernel based on debian's RT-preempt | 15:50 |
| HarryHaaren | right. I'm going to get back to you when the SSD arrives, because my best RT audio experience was on 64Studio 2.1 when it came out! | 15:50 |
| rgareus | HarryHaaren: I used debian's RT-kernel for a while -- until I started writing ALSA drivers. | 15:50 |
| rgareus | HarryHaaren: 64studio. aah, those were the days :) | 15:51 |
| HarryHaaren | they were! I've been waiting for 3.0 for *years* :D | 15:51 |
| rgareus | HarryHaaren: SSD setup is quite orthogonal. has little to do with the kernel. | 15:55 |
| HarryHaaren | rgareus, yep i know. But it gives me a clean slate to do a fresh install, without having to re-do my current developing install. I'll probably install AvLinux: assuming the kernel is -RT and stable | 15:57 |
| rgareus | HarryHaaren: it has a RT kernel. but I don't know if it includes tweaks for SSD | 15:58 |
| rgareus | HarryHaaren: (e.g make /tmp a tmpfs, redirect firefox's cache to /tmp etc, etc) | 15:58 |
| HarryHaaren | that's fine.. i shouldn't be using firefox on stage anyway :D | 16:00 |
| HarryHaaren | I got a sound-techie job doing live sound, so I'm going to be in Germany / Holland area for the LAC time. Might try arrive a bit earlier.. if there's things to do? | 16:01 |
| rgareus | HarryHaaren: swappiness is another one and elevator=noop | 16:01 |
| rgareus | HarryHaaren: you could share a ride with Glen (avlinux) from Holland | 16:02 |
| HarryHaaren | yes, he pinged me about that a few weeks ago: at the time there was no talk of this work yet | 16:02 |
| HarryHaaren | thanks for reminding me, i'll get in touch with him :) | 16:02 |
| rgareus | HarryHaaren: as for stuff to organize @ZKM pre LAC, I don't know, yet. | 16:03 |
| HarryHaaren | rgareus, that's cool, i'll post on LAC-team or something once I know myself what's going on. And if I'm going down with Glen, then perhaps it'll only be for the pre-conf Pizza | 16:04 |
| rgareus | HarryHaaren: I'm planning to do the same, arrive sometime Wednesday afternoon. | 16:05 |
| HarryHaaren | rgareus, awesome. I'm looking forward to it! Going to do a little video for the new ArtyFX distortion plugin now, see you! | 16:07 |
| *** Gethiox has quit IRC | 17:20 | |
| *** Gethiox2 has joined #lv2 | 17:21 | |
| *** awilliams has joined #lv2 | 17:35 | |
| *** rncbc has joined #lv2 | 17:51 | |
| rgareus | drobilla: around? | 18:17 |
| rgareus | drobilla: libsuil's qt4_in_gtk2 does not allow to open multiple QT GUIs at the same time. QApplication is a singleton. | 18:18 |
| rgareus | drobilla: I did a quick test and found a workaround/solution: http://pastebin.com/bx9iEMCU | 18:18 |
| rgareus | drobilla: though I'd like to discuss/clean it up before submitting a patch | 18:18 |
| *** wrl has joined #lv2 | 18:21 | |
| rgareus | drobilla: I did a quick check w/ reference count; but it looks like one must only ever call new QApplication once per process | 18:21 |
| rgareus | rncbc: do you know the semantics of QApplication ? | 18:21 |
| rgareus | rncbc: can one a = new QApplication; delete a; a = new QApplication; without issues? | 18:22 |
| rncbc | rgareus: a guess not. use qApp() to check if an instance is already up | 18:23 |
| rncbc | rgareus: i use something similar on the vee-ones wrt. lv2_external_ui | 18:25 |
| rgareus | rncbc: qApp is a good hint. | 18:27 |
| rncbc | rgareus: i use to check whether qApp is not null and ++refcount on lv2_ui_instantiate; if QApp is null then new QApplication() | 18:30 |
| rncbc | qApp | 18:30 |
| rncbc | rgareus: on lv2_ui_cleanup:: --refcount; when zero => delete QApplication() | 18:31 |
| drobilla | rgareus: Yeah, that part was clearly wrong, but a QApplication was needed somehow IIRC, so.. | 18:31 |
| drobilla | static anything in a dynamically loaded module makes me uneasy | 18:32 |
| drobilla | I duno, do whatever rncbc, I don't know anything about Qt :) | 18:33 |
| rgareus | drobilla: yeah, I've just changed it to use qApp; but that's just a Macro to access QT's static QCoreApplication::instance () | 18:34 |
| drobilla | rgareus: Yeah, just read that bit. | 18:34 |
| drobilla | rgareus: Well, since this should only be used in a Gtk app anyway, it's probably Safe Enough™ | 18:34 |
| rgareus | ref-counting and deleting QApplication instance with the last plugin UI and re-creting it with the next, kind of works | 18:36 |
| rgareus | the probem is QStyle | 18:36 |
| rgareus | the host does not crash but the plugin GUI looks completely odd. | 18:36 |
| rgareus | and I get lots of warnings in the terminal | 18:36 |
| rgareus | the problem is QGtkStyle::QGtkStyle | 18:37 |
| * drobilla finds it odd that none of this has come up before | 18:37 | |
| rgareus | so http://pastebin.com/3q8fVK4k sees to be the right thing to do | 18:38 |
| drobilla | Not a lot of people using Q* plugins in Ardour I suppose | 18:38 |
| rgareus | drobilla: ...or ingen | 18:39 |
| rgareus | drobilla: rui's plugins are external-ui | 18:39 |
| rgareus | drobilla: and I think it's only qmidiarp that currently has a QT gui | 18:39 |
| rgareus | drobilla: shall I submit the patch to your tracker? or can you pick it up from pastebin? | 18:40 |
| drobilla | There's not a lot of people using Ingen period, and it doesn't support external-ui | 18:41 |
| drobilla | oh, right. | 18:41 |
| drobilla | That whole problem. | 18:42 |
| drobilla | rgareus: Does it work? | 18:42 |
| rncbc | rgaureus: my plugins are dual lv2_external_ui and lv2_qt4_ui | 18:43 |
| *** edogawa has joined #lv2 | 18:43 | |
| rgareus | drobilla: yes | 18:44 |
| rgareus | drobilla: it works | 18:44 |
| rgareus | drobilla: tested by me and dharanamrs (qmidiarp's author) | 18:44 |
| rgareus | drobilla: it has the side-effect of keeping the QApplication around until application exit... which is not nice. but I currently see no other solution. | 18:45 |
| rgareus | theoretically one should be able to ref-count it. | 18:46 |
| rgareus | but pratically it takes a QT master to get this right. | 18:46 |
| drobilla | rgareus: can't apply patch from pastebin, mangled somehow. send me directly somehow | 18:47 |
| rgareus | drobilla: will do. | 18:48 |
| drobilla | (this always happens and I don't get why, surely patch is smart enough to deal with line end mangling and such) | 18:48 |
| rgareus | FWIW, here's the ref-count experiment, with asserts() in QGtkStyle::QGtkStyle http://pastebin.com/7CavmFqq | 18:48 |
| rgareus | drobilla: your trac is dead slow :( http://dev.drobilla.net/attachment/ticket/961 | 18:52 |
| drobilla | Yeah, I am not impressed with Dreamhost lately but I don't know what to do about it | 18:57 |
| drobilla | or want to spend the time dealing with it | 18:57 |
| drobilla | rgareus: thanks | 18:59 |
| *** awilliams has quit IRC | 19:01 | |
| rgareus | drobilla: how to you pronounce 'ingen' ? is that a 'g' as in ingenuity ? | 19:02 |
| drobilla | rgareus: It's kind of a pseudo-pun on "engine" and/or "instrument generator", so yeah, soft g | 19:03 |
| *** falktx has joined #lv2 | 19:17 | |
| *** HarryHaaren has quit IRC | 19:27 | |
| *** rncbc has quit IRC | 19:40 | |
| *** mlpug has quit IRC | 20:21 | |
| falktx | drobilla: I tried to build pugl with strict flags for fun | 20:23 |
| falktx | drobilla: here's the needed changes: https://github.com/falkTX/Carla/commit/2a857b0529c7032125b31bb28da669a8ac160602 :) | 20:23 |
| falktx | setModifiers arg was set to long because otherwise all calls to it would need a uint32_t cast | 20:24 |
| falktx | less changes this way | 20:24 |
| drobilla | falktx: I do not use long, and neither should you. | 20:26 |
| falktx | the x11 api does though | 20:26 |
| drobilla | fair enough. | 20:27 |
| drobilla | why the hell does pugl have signed width anyway | 20:28 |
| drobilla | because glut does | 20:28 |
| falktx | I'd welcome a change to that | 20:28 |
| falktx | I use uint internally here, have to cast everytime | 20:28 |
| drobilla | well, puglInit is going to completely change anyway | 20:28 |
| *** awilliams has joined #lv2 | 20:49 | |
| *** Gethiox2 has quit IRC | 21:15 | |
| *** Gethiox2 has joined #lv2 | 21:16 | |
| *** HarryHaaren has joined #lv2 | 21:18 | |
| HarryHaaren | drobilla, can you forsee any backwards-compat issues by moving plugins from individual .so files into one collective .so? (Related to OpenAV ArtyFX) | 21:35 |
| drobilla | HarryHaaren: I don't think that shouls cause any problems | 21:42 |
| HarryHaaren | cool. I'd like to clean up the source & refactor code & build system. Seems cleaner to have a single .so "artyFX" instead of a bunch of files. Thanks! | 21:43 |
| rgareus | HarryHaaren: the problems here are/were not the .so but the huge .ttl | 22:20 |
| drobilla | Yes, that you should not do. | 22:20 |
| rgareus | HarryHaaren: but drobilla did an update to lilv to mitigate the effect | 22:20 |
| drobilla | Though lilv handles it better now. | 22:20 |
| rgareus | I know some of harry's plugins have > 100 ports. | 22:21 |
| HarryHaaren | got it yeah, one manifest.ttl, with "links" to each plugins .ttl: roomy.ttl, ducka.ttl etc | 22:21 |
| HarryHaaren | only Fabla so far.. and that was to allow the automation of the vol/pan/pitch/ADSR, 16x 7 dials really adds up! | 22:22 |
| rgareus | I'm planning to break out various meters into their own .ttl, too. But, for now it's a good lilv performance test :) 62 plugins in 5K lines TTL | 22:25 |
| drobilla | The mode is probably the slow part by a long shot, serd will plow through data as fast as most I/O systems can feed it. | 22:26 |
| drobilla | But it's fast enough to not be noticeable again so I don't care. | 22:26 |
| rgareus | drobilla: the effective size and # of plugins won't shrink. | 22:27 |
| rgareus | drobilla: is the file still scanned N times (once per plugin)? | 22:28 |
| drobilla | rgareus: For hosts that load everything about all plugins at startup, sure. | 22:28 |
| drobilla | That is not the only use case, which is why data is split up. | 22:28 |
| drobilla | (e.g. jalv won't) | 22:29 |
| drobilla | rgareus: No, that was the fix. | 22:29 |
| rgareus | drobilla: well, opening 64 files (or 32 if I group it) instead of just one is also overhead. | 22:29 |
| rgareus | drobilla: the CPU parsing one file repetedly may be faster than disk i/O for small chunks | 22:29 |
| drobilla | rgareus: Unlikely, but irrelevant anyway since the point is to not force hosts to load a bunch of unnecessary data | 22:30 |
| drobilla | It's a bit crap that so many do in the first place since it's often just for a type and label, but whatever. | 22:32 |
| *** edogawa has quit IRC | 23:14 | |
| *** falktx has quit IRC | 23:24 | |
Generated by irclog2html.py 2.13.0 by Marius Gedminas - find it at mg.pov.lt!
